Propagate index cancellation through scan and extraction

TESTING_GUIDE.md

@@ -40,6 +40,8 @@ The test project mirrors the production areas closely.
   End-to-end upgrade path: seeds a pre-column legacy DB, opens it through `TryMigrateForRead`, and exercises the read paths that touch nullable symbol ordinals (outline, symbol search, nearby, unused, analyze bundle) to lock in the real-world failure mode behind #58 / #49.
 - `IndexCommandRunnerTests.cs`, `QueryCommandRunnerTests.cs`, `ProgramCliTests.cs`, `InstallScriptTests.cs`
   CLI parsing, command execution, and installer behavior. `ProgramCliTests.cs` covers top-level entrypoint behavior that must be exercised through a subprocess, while `InstallScriptTests.cs` runs focused bash snippets against `install.sh` in library mode to lock in release-installer regressions without performing real network installs.
+- `IndexCommandRunnerTests.Run_CancelDuringFreshIndex_ReturnsInterruptedJson`, `Run_CancelDuringDryRunScan_ReturnsInterruptedJson`, and `Run_CancelBeforeFreshScan_ReturnsInterruptedJson`
+  exercise the same in-process cancellation paths used after Ctrl-C/SIGINT wiring, including scan-time cancellation, so interrupted index runs keep returning the canonical JSON error contract.
 - `SymbolExtractorTests.Extract_CSharp_InstallScriptFixture_CompletesWithinPracticalBudget`
   is a coarse runaway guard for the real `InstallScriptTests.cs` C# extraction fixture. Its wall-clock budget is intentionally broader than a benchmark so slower or noisy CI hosts do not fail the suite for ordinary variance.
 - `IndexCommandRunnerTests.RunBackfillFold_PublishedTrimmedBinary_SerializesSuccessAndErrorJson`
@@ -225,6 +227,8 @@ dotnet test --filter "FullyQualifiedName~GitHelperTests"
   エンドツーエンドのアップグレード経路: カラム追加前のレガシー DB を用意し、`TryMigrateForRead` 経由で開いてから NULL になりうるシンボル列を触る read path（outline、シンボル検索、近傍、unused、analyze バンドル）を一通り叩き、#58 / #49 の実機失敗モードを固定する。
 - `IndexCommandRunnerTests.cs`、`QueryCommandRunnerTests.cs`、`ProgramCliTests.cs`、`InstallScriptTests.cs`
   CLI の引数解析、コマンド実行、installer 挙動のテスト。`ProgramCliTests.cs` はグローバル引数の解釈や完全な CLI 起動フローのように subprocess 経由で確認すべき Program エントリポイント挙動を扱い、`InstallScriptTests.cs` は `install.sh` を library mode で source した bash snippet を実行して、実ネットワーク install を行わずに release installer の回帰を固定する。
+- `IndexCommandRunnerTests.Run_CancelDuringFreshIndex_ReturnsInterruptedJson`、`Run_CancelDuringDryRunScan_ReturnsInterruptedJson`、`Run_CancelBeforeFreshScan_ReturnsInterruptedJson`
+  Ctrl-C/SIGINT 配線後に使われる in-process cancellation 経路を、scan 中のキャンセルも含めて検証し、interrupted index run が標準の JSON error contract を返し続けることを固定する。
 - `SymbolExtractorTests.Extract_CSharp_InstallScriptFixture_CompletesWithinPracticalBudget`
   は実ファイル `InstallScriptTests.cs` を C# 抽出に通す coarse な runaway guard です。wall-clock の予算は benchmark より意図的に広く取り、遅い / 混雑した CI host で通常の揺れだけにより suite が失敗しないようにしています。
 - `IndexCommandRunnerTests.RunBackfillFold_PublishedTrimmedBinary_SerializesSuccessAndErrorJson`

changelog.d/unreleased/1818.fixed.md

@@ -0,0 +1,19 @@
+---
+category: fixed
+issues:
+  - 1818
+affected:
+  - tests/CodeIndex.Tests/IndexCommandRunnerTests.cs
+  - tests/CodeIndex.Tests/FileIndexerTests.cs
+  - tests/CodeIndex.Tests/SymbolExtractorTests.cs
+  - tests/CodeIndex.Tests/ReferenceExtractorTests.cs
+  - TESTING_GUIDE.md
+---
+
+## English
+
+- **Added cancellation-path regression tests for indexing and extractors (#1818)** — the test suite now exercises in-process index cancellation and cancelled-token entry points for the scanner, symbol extractor, and reference extractor so signal/Ctrl-C wiring regressions are caught earlier.
+
+## 日本語
+
+- **index と抽出器のキャンセル経路の回帰テストを追加しました (#1818)** — テストスイートは in-process index cancellation と、scanner・symbol extractor・reference extractor の cancelled-token 入口を検証し、signal / Ctrl-C 配線の退行をより早く検出します。

changelog.d/unreleased/1841.fixed.md

@@ -0,0 +1,19 @@
+---
+category: fixed
+issues:
+  - 1841
+affected:
+  - src/CodeIndex/Indexer/Scanning/FileIndexer.cs
+  - src/CodeIndex/Indexer/Symbols/SymbolExtractor.cs
+  - src/CodeIndex/Indexer/References/ReferenceExtractor.cs
+  - src/CodeIndex/Cli/IndexCommandRunner.cs
+  - src/CodeIndex/Mcp/McpToolHandlers.cs
+---
+
+## English
+
+- **Index cancellation now reaches scanner, file IO, and extractor work (#1841)** — `cdidx index` and MCP indexing now pass cancellation tokens into file scanning, raw file reads, symbol extraction, and reference extraction so Ctrl-C and request cancellation can stop deeper in-flight work sooner.
+
+## 日本語
+
+- **index のキャンセルが scanner・ファイル IO・抽出処理まで届くようになりました (#1841)** — `cdidx index` と MCP indexing は cancellation token をファイル走査、raw file 読み込み、symbol 抽出、reference 抽出に渡すため、Ctrl-C やリクエストキャンセル後に深い処理もより早く停止できます。

src/CodeIndex/Cli/IndexCommandRunner.cs

@@ -278,14 +278,30 @@ void RecordDryRunScanErrors(IEnumerable<FileIndexer.ScanError> scanErrors)
                 }
             }
 
+            int WriteDryRunInterrupted() => WriteCommandError(
+                options.Json,
+                jsonOptions,
+                "Interrupted before dry-run scan completed.",
+                CommandExitCodes.Interrupted,
+                "Rerun `cdidx index --dry-run` when you are ready to inspect the candidate files again.",
+                CommandErrorCodes.Interrupted);
+
             if (options.UpdateFiles.Count > 0)
             {
                 // --files: only the specified files / --files: 指定ファイルのみ
                 var relevantIgnoreFileChanged = ContainsRelevantIgnoreFileUpdate(options.ProjectPath, options.UpdateFiles);
                 var updatePaths = NormalizeUpdateFileTargets(options.ProjectPath, options.UpdateFiles, options.Json);
                 if (relevantIgnoreFileChanged || ContainsIgnoreFilePath(updatePaths))
                 {
-                    var scanResult = dryIndexer.ScanFilesDetailed();
+                    FileIndexer.ScanFilesResult scanResult;
+                    try
+                    {
+                        scanResult = dryIndexer.ScanFilesDetailed(cancellationToken: indexCancellation.Token);
+                    }
+                    catch (OperationCanceledException) when (indexCancellation.IsCancellationRequested)
+                    {
+                        return WriteDryRunInterrupted();
+                    }
                     dryCandidates = scanResult.Files;
                     RecordDryRunScanErrors(scanResult.Errors);
                 }
@@ -340,7 +356,15 @@ void RecordDryRunScanErrors(IEnumerable<FileIndexer.ScanError> scanErrors)
 
                 if (relevantIgnoreFileChanged || ContainsIgnoreFilePath(changedFiles))
                 {
-                    var scanResult = dryIndexer.ScanFilesDetailed();
+                    FileIndexer.ScanFilesResult scanResult;
+                    try
+                    {
+                        scanResult = dryIndexer.ScanFilesDetailed(cancellationToken: indexCancellation.Token);
+                    }
+                    catch (OperationCanceledException) when (indexCancellation.IsCancellationRequested)
+                    {
+                        return WriteDryRunInterrupted();
+                    }
                     dryCandidates = scanResult.Files;
                     RecordDryRunScanErrors(scanResult.Errors);
                 }
@@ -354,7 +378,15 @@ void RecordDryRunScanErrors(IEnumerable<FileIndexer.ScanError> scanErrors)
             }
             else
             {
-                var scanResult = dryIndexer.ScanFilesDetailed();
+                FileIndexer.ScanFilesResult scanResult;
+                try
+                {
+                    scanResult = dryIndexer.ScanFilesDetailed(cancellationToken: indexCancellation.Token);
+                }
+                catch (OperationCanceledException) when (indexCancellation.IsCancellationRequested)
+                {
+                    return WriteDryRunInterrupted();
+                }
                 dryCandidates = scanResult.Files;
                 RecordDryRunScanErrors(scanResult.Errors);
             }
@@ -1706,7 +1738,7 @@ void ThrowIfUpdateCancelled()
             var expandHeartbeat = StartJsonPhaseHeartbeat("expanding C# update set for static interface contracts");
             try
             {
-                foreach (var filePath in indexer.ScanFilesDetailed().Files)
+                foreach (var filePath in indexer.ScanFilesDetailed(cancellationToken: cancellationToken).Files)
                 {
                     var detection = FileIndexer.TryDetectLanguage(filePath);
                     if (detection.Status == FileIndexer.FileProbeStatus.Supported
@@ -2018,7 +2050,7 @@ void ThrowIfUpdateCancelled()
                     continue;
                 }
 
-                var (record, content, rawBytes, warning) = indexer.BuildRecordWithRawBytes(absPath);
+                var (record, content, rawBytes, warning) = indexer.BuildRecordWithRawBytes(absPath, cancellationToken);
 
                 if (warning != null && !options.Json && !options.Quiet)
                 {
@@ -2080,7 +2112,7 @@ void ThrowIfUpdateCancelled()
                 var chunks = ChunkSplitter.Split(fileId, content);
                 writer.InsertChunks(chunks);
                 currentUpdatePath = FormatIndexPhasePath(relPath, "symbols");
-                var symbols = SymbolExtractor.Extract(fileId, record.Lang, content, absPath, Path.GetFullPath(options.ProjectPath!));
+                var symbols = SymbolExtractor.Extract(fileId, record.Lang, content, absPath, Path.GetFullPath(options.ProjectPath!), cancellationToken);
                 SymbolExtractor.ApplyFamilyScope(symbols, indexer.GetFamilyScopeKey(absPath, record.Lang));
                 var fileContext = new FileContext(projectRoot, record.Path, absPath, record.Lang);
                 postExtractionHooks.OnSymbolsExtracted(fileContext, symbols);
@@ -2094,7 +2126,8 @@ void ThrowIfUpdateCancelled()
                     content,
                     symbols,
                     record.Path,
-                    record.Lang == "csharp" ? csharpWorkspace.Symbols : null);
+                    record.Lang == "csharp" ? csharpWorkspace.Symbols : null,
+                    cancellationToken);
                 postExtractionHooks.OnReferencesExtracted(fileContext, references);
                 writer.InsertReferences(references);
                 // Validate content for encoding issues / エンコーディング問題を検証
@@ -3179,9 +3212,13 @@ void ThrowIfFullScanCancelled(int filesProcessed, int? filesTotal)
         try
         {
             ThrowIfFullScanCancelled(0, null);
-            scanResult = indexer.ScanFilesDetailed(checkpointedDirectories, continueOnError: true);
+            scanResult = indexer.ScanFilesDetailed(checkpointedDirectories, continueOnError: true, cancellationToken: cancellationToken);
             ThrowIfFullScanCancelled(0, null);
         }
+        catch (OperationCanceledException) when (cancellationToken.IsCancellationRequested)
+        {
+            throw new IndexInterruptedException(0, null);
+        }
         finally
         {
             StopJsonPhaseHeartbeat(scanHeartbeat);
@@ -3508,7 +3545,7 @@ void StopJsonHeartbeat()
                         {
                             var relativeFilePath = FileIndexer.NormalizePathSeparators(Path.GetRelativePath(projectRoot, filePath));
                             activeJsonExtractionPhases[workerIndex] = FormatIndexPhasePath(relativeFilePath, "reading");
-                            var (record, content, rawBytes, warning) = indexer.BuildRecordWithRawBytes(filePath);
+                            var (record, content, rawBytes, warning) = indexer.BuildRecordWithRawBytes(filePath, cancellationToken);
                             IReadOnlyList<ChunkRecord>? chunks = null;
                             IReadOnlyList<SymbolRecord>? symbols = null;
                             IReadOnlyList<ReferenceRecord>? references = null;
@@ -3518,7 +3555,7 @@ void StopJsonHeartbeat()
                                 activeJsonExtractionPhases[workerIndex] = FormatIndexPhasePath(record.Path, "chunking");
                                 chunks = ChunkSplitter.Split(0, content);
                                 activeJsonExtractionPhases[workerIndex] = FormatIndexPhasePath(record.Path, "symbols");
-                                symbols = SymbolExtractor.Extract(0, record.Lang, content, filePath, Path.GetFullPath(options.ProjectPath!));
+                                symbols = SymbolExtractor.Extract(0, record.Lang, content, filePath, Path.GetFullPath(options.ProjectPath!), cancellationToken);
                                 SymbolExtractor.ApplyFamilyScope(symbols, indexer.GetFamilyScopeKey(filePath, record.Lang));
                                 activeJsonExtractionPhases[workerIndex] = FormatIndexPhasePath(record.Path, "references");
                                 references = ReferenceExtractor.Extract(
@@ -3527,7 +3564,8 @@ void StopJsonHeartbeat()
                                     content,
                                     symbols,
                                     record.Path,
-                                    record.Lang == "csharp" ? csharpWorkspace.Symbols : null);
+                                    record.Lang == "csharp" ? csharpWorkspace.Symbols : null,
+                                    cancellationToken);
                                 activeJsonExtractionPhases[workerIndex] = FormatIndexPhasePath(record.Path, "validating");
                                 issues = FileIndexer.ValidateContent(record.Path, rawBytes, content);
                             }
@@ -3698,7 +3736,7 @@ void StopJsonHeartbeat()
                     writer.InsertChunks(chunks);
                     currentJsonIndexFile = FormatIndexPhasePath(record.Path, "symbols");
                     var symbols = item.Symbols == null
-                        ? SymbolExtractor.Extract(fileId, record.Lang, item.Content!, item.FilePath, Path.GetFullPath(options.ProjectPath!))
+                        ? SymbolExtractor.Extract(fileId, record.Lang, item.Content!, item.FilePath, Path.GetFullPath(options.ProjectPath!), cancellationToken)
                         : ReassignSymbolFileIds(item.Symbols, fileId);
                     if (item.Symbols == null)
                         SymbolExtractor.ApplyFamilyScope(symbols, indexer.GetFamilyScopeKey(item.FilePath, record.Lang));
@@ -3717,7 +3755,8 @@ void StopJsonHeartbeat()
                             item.Content!,
                             symbols,
                             record.Path,
-                            record.Lang == "csharp" ? csharpWorkspace.Symbols : null)
+                            record.Lang == "csharp" ? csharpWorkspace.Symbols : null,
+                            cancellationToken)
                         : ReassignReferenceFileIds(item.References, fileId);
                     postExtractionHooks.OnReferencesExtracted(fileContext, AsMutableList(references));
                     writer.InsertReferences(references);
@@ -4289,14 +4328,14 @@ private static CSharpStaticInterfaceWorkspaceSymbols BuildCSharpStaticInterfaceW
             try
             {
                 reportCurrentFile?.Invoke(relativePath);
-                var (record, content, _, _) = indexer.BuildRecordWithRawBytes(absolutePath);
+                var (record, content, _, _) = indexer.BuildRecordWithRawBytes(absolutePath, cancellationToken);
                 if (record.Lang != "csharp")
                     continue;
 
                 if (!MayContainCSharpStaticInterfaceContract(content))
                     continue;
 
-                pendingSymbols.AddRange(SymbolExtractor.Extract(0, record.Lang, content, record.Path));
+                pendingSymbols.AddRange(SymbolExtractor.Extract(0, record.Lang, content, record.Path, cancellationToken: cancellationToken));
             }
             catch (Exception ex) when (ex is IOException or UnauthorizedAccessException or InvalidOperationException)
             {

src/CodeIndex/Indexer/References/ReferenceExtractionContext.cs

@@ -13,4 +13,5 @@ public sealed record ReferenceExtractionContext(
     IReadOnlyList<SymbolRecord> Symbols,
     string? Path = null,
     IReadOnlyList<SymbolRecord>? WorkspaceSymbols = null,
-    string? RequestedLanguage = null);
+    string? RequestedLanguage = null,
+    CancellationToken CancellationToken = default);

src/CodeIndex/Indexer/References/ReferenceExtractor.cs

@@ -908,8 +908,10 @@ public static List<ReferenceRecord> Extract(
         string content,
         IReadOnlyList<SymbolRecord> symbols,
         string? path = null,
-        IReadOnlyList<SymbolRecord>? workspaceSymbols = null)
+        IReadOnlyList<SymbolRecord>? workspaceSymbols = null,
+        CancellationToken cancellationToken = default)
     {
+        cancellationToken.ThrowIfCancellationRequested();
         var requestedLanguage = lang;
         var pluginLanguage = NormalizePluginLanguage(lang);
         if (!TryGetExtractor(lang, out var extractor))
@@ -924,6 +926,7 @@ public static List<ReferenceRecord> Extract(
             if (content.Contains('\r'))
                 content = content.Replace("\r\n", "\n").Replace("\r", "\n");
             content = FileIndexer.StripLineLeadingInvisibles(content);
+            cancellationToken.ThrowIfCancellationRequested();
 
             return pluginExtractor.Extract(
                     fileId,
@@ -941,11 +944,13 @@ public static List<ReferenceRecord> Extract(
             symbols,
             path,
             workspaceSymbols,
-            requestedLanguage));
+            requestedLanguage,
+            cancellationToken));
     }
 
     internal static List<ReferenceRecord> ExtractCore(ReferenceExtractionContext request)
     {
+        request.CancellationToken.ThrowIfCancellationRequested();
         var fileId = request.FileId;
         var language = request.Language;
         var content = request.Content;
@@ -958,6 +963,7 @@ internal static List<ReferenceRecord> ExtractCore(ReferenceExtractionContext req
 
         if (!TryPrepareReferenceLines(language, content, isRazorFile, out var preparedInput))
             return [];
+        request.CancellationToken.ThrowIfCancellationRequested();
 
         content = preparedInput.Content;
         var lines = preparedInput.Lines;
@@ -1167,6 +1173,9 @@ bool HasActiveSameFileCSharpTypeCandidate(string typeExpression, int lineNumber)
 
         for (int i = 0; i < lines.Length; i++)
         {
+            if ((i & 0x3f) == 0)
+                request.CancellationToken.ThrowIfCancellationRequested();
+
             var lineNumber = i + 1;
             var originalLine = lines[i];
             var preparedLine = luaPreparedLines?[i] ?? lispReferenceLines?[i] ?? preparedLines[i];